Corsair CX550 issue with UPS

[whitestar_999]

UPS batteries are working fine with FSP with same system load that shuts down CX550 with UPS overload beep(playing a video file with madvr that loads ~65% of 1050Ti as per gpu-z). So the only explanations left are: a)high inrush current or b)degraded transfer time of UPS(14 months old) less than CX550 but more than FSP Saga II.

@gta5 CX550 worked fine with inverter when I checked it for 2 days but currently I am again using FSP with UPS.Because it is summer so voltage fluctuation is quite high sometimes during day making UPS constantly making clicking sounds at a duration of 1-2 seconds(I then turn off power to UPS for 2-3 minutes & let system run on UPS till fluctuations are gone).I am not sure it is a good idea to directly connect system to inverter without UPS at those times.

 

[rijinpk1]

^^which media player do you use and what settings? I just tried madvr with media player classic now(I am using madvr for the first time) and played a 4K video. GPU is being utilized and usage was 14%-35% with 14% majority of the time. I have corsair cx450 and radeon HD 7770 by the way.

 

[whitestar_999]

What resolution screen are you using?Madvr's most taxing algorithms are for upscaling so if you try to play a 480p/720p 25/30fps video on a 1080p screen using high quality settings(removing compression artifacts & NGU for upscaling) then it should easily load 1050Ti/similar card to ~70-80%. For 60fps 720p/1080p video on a 4k screen even a 1070 will not be enough at high quality settings.

P.S.I am playing a 13 years old 640*480p avi xvid video file on 1080p screen using madvr with pot player(not relevant as load on gfx card remains almost same with mpc-hc too).

 

[rijinpk1]

1600 x 900, it is a 20" LG monitor.

 

[whitestar_999]

That's why on doom9 forum I have seen people complaining about even 1080 not able to run smoothly some videos at highest quality madvr settings.

 

[whitestar_999]

@rijinpk1 can you do a simple test if possible? run mpc-hc with madvr & a high quality 720p video(bitrate 4-5mbps) & then use NGU(sharp) in madvr upscaling algorithm while simultaneously running some cpu benchmark(like prime 95) if gpu-z show 7770 load below 60-70%.Then turn off the mains power to UPS(which brand is yours & how old are its batteries) for 2-3 seconds & see if your system stays on.

 

[rijinpk1]

will do it tomorrow and revert.

 

[rijinpk1]

I switched off the mains for almost 10 seconds and system was working fine. I am using V-guard slender plus 600VA(300W) UPS which is 6 years old and battery was changed about 7-8 months ago.

 

[whitestar_999]

Thanks!Are you using CX450 after getting new batteries or from before that?My issue is that UPS works fine with another 400W psu(non activePFC) & can easily give backup of at least a minute with heavy system load but with CX550 it just overloads with any power cut.I am not sure if even changing batteries will help,it seems either issue is with UPS or psu.

 

[chimera201]

My theory is that your UPS's total transfer time is between 12ms and 21ms. CX550's holdup is 11.93ms and FSP's is 21.6ms. FSP has enough holdup time to support the transfer. Even higher than the 50Hz cycle (20ms). Either you have to get a UPS with less than 10ms transfer time or a PSU with >20ms holdup time.

 

[whitestar_999]

I understand this point but then why UPS is still able to work fine with CX550 under low load condition(aka just downloading in background with nothing else).Is it because transfer time required by CX550 depends on load at the time of power cut?Also there was no issue in the first 3-4 days of using CX550 with UPS & there were power cuts in that period incl when system was under moderate load.

 

[chimera201]

I understand this point but then why UPS is still able to work fine with CX550 under low load condition(aka just downloading in background with nothing else).Is it because transfer time required by CX550 depends on load at the time of power cut?Also there was no issue in the first 3-4 days of using CX550 with UPS & there were power cuts in that period incl when system was under moderate load.

That is as explained by gta5 that the capacitors would discharge faster under load. Your UPS's transfer time is probably cutting very close to the CX550's holdup. Under load it exceeded it. As for the first few days it worked fine, it could be that the UPS or PSU degraded later. More likely the PSU since it was taking in the simulated sine wave.

 

[whitestar_999]

I heard something similar on jonnyguru forums too.Is it possible to know/measure this "degradation" of psu or UPS?I mean if psu is "degraded" does this mean CX550 transfer time is now even less than 11.93ms or does this mean that my UPS transfer time is now more than 11.93ms(my UPS is iball 600VA around 14 months old with stated transfer time of 4-8ms).Is it possible to get UPS repaired to improve transfer time?

 

[chimera201]

I don't think they will repair it to improve it's transfer time. In India repair happens when something is broken. But your UPS is functional for them since they will test it on their own PSUs that has enough holdup. I would suggest getting an APC one.

Here this guy says the same thing:
Corsair CX550M – farewell group design

^ He also fixes it in the conclusion.

 

[gta5]

1) again, it is meaningless to consider Hold up time without taking into consideration the

"System load"

hold up time is not static/fixed, it is totally dependent on load

Just like you cannot talk about battery life/runtime of smartphone without taking into account load/usage pattern.. You can run the same phone for days on standby or you can deplete the battery in just 3 hours or so if you play intenisve game continuously..

Or

From your UPS , You can get backup time of only 5-10 minutes if you run your pc under heavy load or you can get more than an hour of backup time from same ups if you only power a bulb or something

The primary cap of PSU works like a mini battery that holds charge..

"Both" the tests given in those links are measuring hold up time at
"100% load of PSU"

Whitestar is not even running his PSU at more than 30% load..

Here , check this graph of a typical PSU of 250 watts

See how hold up time varies in relation to load a lotttt..

At 250 watts it has 25ms or so , at 150 watt it is 50ms , at 50 watts that is 140 ms or so

*www.artesyn.com/power/assets/reference/appnotes/an55.pdf

For whitestar system load , his hold up time is easily more than 25-30 ms .. So this is why hold up time highlyyyy unlikely to be the reason..

2) Afaik transfer time of UPS is not mainly dependant on load .. It is mainly dependant at which point on AC sinecurve cycle does the power is lost.. But in any case genrally a decent UPS will not have more than 12 ms transfer time

His UPS works fine at low system load and also under load with his other psu suggesting transfer time is not the problem.. Unless you think that somehow transfer time has "coincidentely" degraded "so much" and landed just right around 30ms or more .. More than cx but less than fsp ( 50ms roughly 400/150x20 )..

Possible but very very less chances of that coincidence.. Even then it would be UPS at fault

3) his PSU under load works fine with his Inverter ..

4) rijin's, quiky008 PSU works fine with roughly similar hold up times with their UPS

Hold up time is highlyy unlikelyyy to be the reason thus..

Will post more later

 

[whitestar_999]

@chimera201 see post#28 @rijinpk1 did a test on my request & his UPS is only 300W(as per specifications but his transfer time is stated as 4ms) while his system draws more power(an i3 vs pentium dual core & a 7770 vs 1050Ti) with a CX450.I just want to be clear about it before buying a new UPS in future for my next system upgrade as psu will be CX550.

@gta5 if @rijinpk1 can use his CX450 with a 300W UPS with more system load than mine then why do you think it still might not work with my CX550 if I get a new 600VA APC UPS.

 

[rijinpk1]

Thanks!Are you using CX450 after getting new batteries or from before that?My issue is that UPS works fine with another 400W psu(non activePFC) & can easily give backup of at least a minute with heavy system load but with CX550 it just overloads with any power cut.I am not sure if even changing batteries will help,it seems either issue is with UPS or psu.

PSU is only about 3-4 months old. you may have to try some trial and error method, I suppose- with another UPS,another active PFC PSU.

 

[rijinpk1]

I heard something similar on jonnyguru forums too.Is it possible to know/measure this "degradation" of psu or UPS?I mean if psu is "degraded" does this mean CX550 transfer time is now even less than 11.93ms or does this mean that my UPS transfer time is now more than 11.93ms(my UPS is iball 600VA around 14 months old with stated transfer time of 4-8ms).Is it possible to get UPS repaired to improve transfer time?

hold up time is measured at 80% load afaik. even if you stress your system components to 100%, you are not going to use the half the power the PSU can supply. 1050 ti doesn't even need a pcie power connector, right?

 

[chimera201]

@gta5 I understand the holdup time is not fixed. I never considered it to be fixed. But the thing is holdup is about capacitors keeping its charge. And inrush happens if the charge is getting depleted.

You bolster the inrush only theory by saying the charge will get depleted faster if "under load" causing inrush.
And you refute the holdup theory by saying the charge will not get depleted under "low load".

^ Both the above statements are contradicting each other.
And the "holdup to load" relation is exponential/logarithmic not linear. It will be very high under very low loads like <50W. But it will quickly drop to 10-20ms when the load is increased (>50%). The problem here is a combination of the UPS having a high transfer time as well as the PSU having a low holdup. The PSU caps might have been degraded as well.

 

[gta5]

If you weren't considering it fixed , then taking 11ms figure for whitestars system load is a bigg error.. That is what I was pointing at.. That figure was measured when PSU was outputting 550 watts in both the tests.. Hence taking that same figure for 150 watt outputt will lead to wrong conclusion

And you refute the holdup theory by saying the charge will not get depleted under "low load".

I didn't say this ..

This is what I said

At low loads , inrush current isn't high to trigger overload condition of UPS , plus capacitors will be less discharged because of low load at idle and will draw less inrush current
Till the time ups resumes

But at high loads , primary capacitors would discharge faster , resulting in more inrush current on top of high PC load

Primary Capacitor will also get discharged at low load , since it will take some minimum 4-6ms for UPS to switch regardless of load.. But since load is low it will get discharged very less in the same amount of time and thus less inrush current plus less system load and Not 0 inrush current..

That less inrush current at low load is well within the limits of UPS , hence it works with low load..

And the "holdup to load" relation is exponential/logarithmic not linear. It will be very high under very low loads like <50W. But it will quickly drop to 10-20ms

Yes , at every 50% of load , hold up time would roughly get double

Let's take 10ms as an example..
So,
550 watts - 10ms
275 watts -20ms
137.5 watts - 40ms
68 watts - 80ms
34 watts - 160ms and so on..

Like I mentioned before, In reality it will not be exactly like it but "roughly" in the same ballpark due to other factors involved..

Check the link of that 250 watts power supply again.. And look at that graph.. Here

250 watts - 27ms
200 watts - 35 ms
150 watts - 47ms
100 watts -72ms
50 watts - 145 ms

Roughly it will follow similar pattern

So for whitestar load even if you take 8 ms figure as worst case at 100% load , his hold up time at 150 watts should easily be more than 25ms at the minimum .. 550/150 x 8ms.. Which is way more than required

So you see why hold up time is highllllllyyy unlikely to be the cause ,